RTOp : Vector Reduction Transformation Operators
Version 1.0 in Trilinos 6.0
This doxygen collection contains documentation to RTOp (reduction/transformation operators) which provides the basic mechanism for implementing vector operations in a flexible and efficient manner.
-
Base interfaces and related code: By far the most important members in this collection is the templated abstract interface for vector reduction/transformation operators
RTOpPack::RTOpT, the abstract reduction object interface RTOpPack::ReductTarget. The types RTOpPack::SubVectorT and RTOpPack::MutableSubVectorT represent chunks of non-const and const vector data presented to RTOpPack::RTOpT::apply_op().
-
Support code: Several different types of support code is also provided.
-
MPI Support: The templated functions
RTOpPack::MPI_apply_op() help handle MPI communication when using RTOp. These function rely on the traits classes Teuchos::RawMPITraits and Teuchos::PrimitiveTypeTraits to perform all of the marshaling and unmarshaling of data. The function RTOp_parallel_calc_overlap() is used by vector subclasses to figure out what processors have data associated with a particular RTOp application.
-
Support for creating concrete RTOPPack::RTOpT subclasses: Several node subclasses and utility macros are provided to help in creating concrete implementations of
RTOpPack::RTOpT and these are documented here.
-
Support for client/server and master/slave configurations: The class
RTOpPack::RTOpServer stores and returns Teuchos::AbstractFactory objects for creating RTOpPack::RTOpT objects just given a name.
-
Library of RTOp subclasses: Many different RTOp subclasses are also provided. While this list is not meant to be (nor does it need to be) exhaustive, it does meet the needs of many different types of ANAs.
Below are links to other related documentation:
Generated on Thu Sep 18 12:39:44 2008 for RTOp : Vector Reduction/Transformation Operators by
1.3.9.1